HYBRID NEURO-MATHEMATICAL MODEL FOR DETECTING INFORMATION THREATS: INTEGRATING LARGE LANGUAGE MODELS AND MODIFIED DIFFUSION PROCESSES
DOI:
https://doi.org/10.26906/SUNZ.2026.3.115Keywords:
information security, information operations, large language models, natural language processing, social network analysis, hybrid neuro-mathematical model, diffusion processes, stochastic modelsAbstract
Background. In the context of the rapid evolution of cyber threats, where destructive narratives spread with unprecedented speed, traditional isolated monitoring systems lose their effectiveness. Subject of research: analytical systems for detecting coordinated information operations in modern cognitive security. Purpose of research: to develop a unified neuro-mathematical architecture that maps non-linear semantic features onto differential equations of spatio-temporal diffusion. Research objectives. To construct a hybrid model based on a large language model, transform semantic outputs into continuous kinematic parameters, and empirically validate the modelʼs efficacy during real information attacks. Research methods. A mixed-methods empirical approach was employed, integrating quantitative mathematical modeling with computational linguistics, the Llama-3.1-8B-Instruct model utilizing the Adaptive Self-Consistency algorithm, and modifying the Bass diffusion model alongside the Neural Hawkes process. Research results. The implementation of thermodynamic logit scaling reduced semantic classification entropy to 0.160 bits. Integrating continuous kinematic parameters into differential equations allowed the system to identify the phase transition of an attack 3 to 4 hours prior to its statistical peak. It was revealed that granular emotional subtype classification without fine-tuning leads to mathematical collapse, entropy surging to 1.349 bits. Conclusions. The research expands the Bass diffusion theory and Hawkes processes by replacing static constants with dynamic, AI-derived forcing functions. It mathematically proves the necessity of macropolarization in feature spaces for stable differential modeling without network fine-tuning.Downloads
References
1. Parmar, M. (2026), “Safety, security, and cognitive risks in world models”, arXiv, arXiv.2604.01346, doi: https://doi.org/10.48550/arXiv.2604.01346
2. Li, Y., Yang, Y., and Huang, H. (2025), “Advances in social bots [Editorial]”, Electronics, vol. 14, no. 24, 4885, doi: https://doi.org/10.3390/electronics14244885 DOI: https://doi.org/10.3390/electronics14244885
3. Sadeghi, M., Dimitriadis, D., Padovese, V., Pozzi, G., Badilini, S., Vercellone, C., Huet, N., Fishman, Z., Pfaller, L., and Adams, N. (2026), “Tracking AI-enabled misinformation: 3,749 AI content farm sites (and counting), plus the top false claims generated by artificial intelligence tools”, NewsGuard, available at: https://www.newsguardtech.com/special-reports/aitracking-center
4. Sabri, M. I. F., Hettiarachchi, H., and Ranasinghe, T. (2025), “Detecting fake news in the era of language models”, Proceedings of the 15th International Conference on Recent Advances in Natural Language Processing, pp. 1036–1043, available at: https://aclanthology.org/2025.ranlp-1.119/ DOI: https://doi.org/10.26615/978-954-452-098-4-119
5. Burghardt, K., Chen, K., and Lerman, K. (2024), “Large language models reveal information operation goals, tactics, and narrative frames”, Proceedings of the ICWSM Workshops, doi: https://doi.org/10.48550/arXiv.2405.03688
6. Stanovov, V., Grabljevec, S., Akhmedova, S., Semenkin, E., Stojanović, R., Rozman, Č., and Škraba, A. (2022), “Identification of COVID-19 spread mechanisms based on first-wave data, simulation models, and evolutionary algorithms”, PLOS One, doi: https://doi.org/10.1371/journal.pone.0279427 DOI: https://doi.org/10.1371/journal.pone.0279427
7. Whiteside, R. D. (2025), “Narrative manipulation, malinfluence operations, and cognitive warfare through large language model poisoning with adversarial noise”, Military Intelligence Professional Bulletin, available at: https://mipb.ikn.army.mil/
8. Ezzeddine, F., Ayoub, O., Giordano, S., Nogara, G., Sbeity, I., Ferrara, E., and Luceri, L. (2023), “Exposing influence campaigns in the age of LLMs: A behavioral-based AI approach to detecting state-sponsored trolls”, EPJ Data Science, vol. 12, no. 1, 46, doi: https://doi.org/10.1140/epjds/s13688-023-00423-4 DOI: https://doi.org/10.1140/epjds/s13688-023-00423-4
9. Cosguner, K., and Seetharaman, P. B. (2022), “Dynamic pricing for new products using a utility-based generalization of the Bass diffusion model”, Management Science, vol. 68, no. 3, pp. 1904–1922, doi: https://doi.org/10.1287/mnsc.2021.4257 DOI: https://doi.org/10.1287/mnsc.2021.4257
10. Liang, L. (2021), “Quantitative analysis of diffusion of public opinion by the modified Bass model for the dyeing steamed bread incident in China”, Saudi Journal of Business and Management Studies, vol. 6, no. 1, pp. 24–27, doi: https://doi.org/10.36348/sjbms.2021.v06i01.004 DOI: https://doi.org/10.36348/sjbms.2021.v06i01.004
11. Ask, T. F., Sütterlin, S., Müller, L., Lugo, R., Østby, L., Jensen, M., Nordheim, K., Gjøvik, J., and Lervik, E. (2025), “Cognitive security: The study and practice of protecting the human mind and other cognitive assets from cognitive threats”, OSF, doi: https://doi.org/10.31234/osf.io/2ftqc_v1 DOI: https://doi.org/10.31234/osf.io/2ftqc_v1
12. Stella, M., Ferrara, E., and De Domenico, M. (2018), “Bots increase exposure to negative and inflammatory content in online social systems”, Proceedings of the National Academy of Sciences, vol. 115, no. 49, pp. 12435–12440, doi: https://doi.org/10.1073/pnas.1803470115 DOI: https://doi.org/10.1073/pnas.1803470115
13. Munir, B., Abbasi, M. Z., Wilson, W. B., and Colombo, A., Jr. (2025), “Evaluating AI in legal operations: A comparative analysis of accuracy, completeness, and hallucinations in ChatGPT-4, Copilot, DeepSeek, Lexis+ AI, and Llama 3”, International Journal of Legal Information, vol. 53, no. 2, pp. 103–114, doi: https://doi.org/10.1017/jli.2025.10052 DOI: https://doi.org/10.1017/jli.2025.10052
14. Barbera, I. (2025), AI privacy risks & mitigations - large language models (LLMs). Practical guide for conducting risk assessments to mitigate privacy and data protection risks in systems using LLMs, European Data Protection Board Support Pool of Experts, available at: https://www.edpb.europa.eu/system/files/2025-04/ai-privacy-risks-and-mitigations-in-llms.pdf
15. Vassilev, A., Oprea, A., Fordyce, A., Anderson, H., Davies, X., and Hamin, M. (2025), Adversarial machine learning: A taxonomy and terminology of attacks and mitigations (NIST AI 100-2e2025), National Institute of Standards and Technology, doi: https://doi.org/10.6028/NIST.AI.100-2e2025 DOI: https://doi.org/10.6028/NIST.AI.100-2e2023
16. Jeong, E., and Choi, Y. S. (2025), “ACR: Adaptive confidence re-scoring for reliable answer selection among multiple candidates”, Applied Sciences, vol. 15, no. 17, 9587, doi: https://doi.org/10.3390/app15179587 DOI: https://doi.org/10.3390/app15179587
17. Alcaraz, A. (2023), Achieving greater self-consistency in large language models, Towards Data Science, available at: https://towardsdatascience.com/achieving-greater-self-consistency-in-large-language-models-6e6cb5f3c5b7
18. Xie, Z., Guo, J., Yu, T., and Li, S. (2024), Calibrating reasoning in language models with internal consistency, NeurIPS 2024, available at: https://neurips.cc/virtual/2024/poster/93260
19. Taherinezhad, F., Momeni Nezhad, M. J., Karimi, S., et. al (2026), “Large language model adaptation strategies in speechbased cognitive screening: Systematic evaluation”, JMIR AI, vol. 5, e82608, doi: https://doi.org/10.2196/82608 DOI: https://doi.org/10.2196/82608
20. Bass, F. M., Krishnan, T. V., and Jain, D. C. (1994), “Why the Bass model fits without decision variables”, Marketing Science, vol. 13, no. 3, pp. 203–223, doi: https://doi.org/10.1287/mksc.13.3.203 DOI: https://doi.org/10.1287/mksc.13.3.203
21. Manshadi, V., Johari, R., and Misra, S. (2015), A modified Bass model for product growth in social networks, available at: https://www-2.rotman.utoronto.ca/MSOM-2015/Documents/MSOM2015_0325.pdf
22. Baek, J., Farias, V. F., Georgescu, A., Levi, R., Peng, T., Sinha, D., Wilde, J. T., and Zheng, A. T. (2025), “The limits to learning a diffusion model”, Management Science, doi: https://doi.org/10.1287/mnsc.2022.02953 DOI: https://doi.org/10.1287/mnsc.2022.02953
23. Laub, P. J., Lee, Y., Pollett, P. K., and Taimre, T. (2025), “Hawkes models and their applications”, Annual Review of Statistics and Its Application, vol. 12, pp. 233–258, doi: https://doi.org/10.1146/annurev-statistics-112723-034304 DOI: https://doi.org/10.1146/annurev-statistics-112723-034304
24. Li, J., Yang, Y., Zhang, Y., Hu, Q., Zhao, A., and Gao, H. (2025), “Public opinion field effect and Hawkes process join hands for information popularity prediction”, Proceedings of the AAAI Conference on Artificial Intelligence, vol. 39, no. 11, pp. 12076–12083, doi: https://doi.org/10.1609/aaai.v39i11.33315 DOI: https://doi.org/10.1609/aaai.v39i11.33315
25. Okawa, M., Iwata, T., Tanaka, Y., Toda, H., Kurashima, T., and Kashima, H. (2021), “Dynamic Hawkes processes for discovering time-evolving communitiesʼ states behind diffusion processes”, Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery & Data Mining, pp. 1276–1286, doi: https://doi.org/10.1145/3447548.3467248 DOI: https://doi.org/10.1145/3447548.3467248
26. Zeng, F., and Gao, W. (2022), “Early rumor detection using neural Hawkes process with a new benchmark dataset”, Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics, pp. 4105–4117, doi: https://doi.org/10.18653/v1/2022.naacl-main.302 DOI: https://doi.org/10.18653/v1/2022.naacl-main.302
27. Kuntur, S., Wróblewska, A., Paprzycki, M., and Ganzha, M. (2025), “Under the influence: A survey of large language models in fake news detection”, IEEE Transactions on AI, vol. 6, no. 2, pp. 458–476, doi: https://doi.org/10.1109/TAI.2024.3471735 DOI: https://doi.org/10.1109/TAI.2024.3471735
28. Eryarsoy, E., Delen, D., Davazdahemami, B., and Topuz, K. (2021), “A novel diffusion-based model for estimating cases, and fatalities in epidemics: The case of COVID-19”, Journal of Business Research, vol. 124, pp. 163–178, doi: https://doi.org/10.1016/j.jbusres.2020.11.054 DOI: https://doi.org/10.1016/j.jbusres.2020.11.054
Downloads
Published
Issue
Section
License
Copyright (c) 2026 Yaroslav Lashyn, Oleksandr Voitko, Vitalii Fedoriienko
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.
